Earth and Planetary Science Letters 321-322 (2012) 20–31

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters

journal homepage: www.elsevier.com/locate/epsl

Geochemical signatures and magmatic stability of terrestrial impact produced zircon

Matthew M. Wielicki ⁎, T. Mark Harrison, Axel K. Schmitt

Department of Earth and Space Sciences, University of California Los Angeles, Los Angeles, CA 90095, USA article info abstract

Article history: Understanding the role of impacts on early Earth has major implications to near surface conditions, but the Accepted 9 January 2012 apparent lack of preserved terrestrial craters >2 Ga does not allow a direct sampling of such events. Ion Available online 1 February 2012 microprobe U–Pb ages, REE abundances and Ti-in-zircon thermometry for impact produced zircon are reported here. These results from terrestrial impactites, ranging in age from ~35 Ma to ~2 Ga, are compared Editor: R.W. Carlson with the detrital Hadean zircon population from Western Australia. Such comparisons may provide the only terrestrial constraints on the role of impacts during the Hadean and early Archean, a time predicted Keywords: ﬂ zircon to have a high bolide ux. Ti-in-zircon thermometry indicates an average of 773 °C for impact-produced zir- early Earth con, ~100 °C higher than the average for Hadean zircon crystals. The agreement between whole-rock based impact zircon saturation temperatures for impactites and Ti-in-zircon thermometry (at aTiO2 =1) implies that Hadean Ti-in-zircon thermometry record actual crystallization temperatures for impact melts. Zircon saturation Late Heavy Bombardment modeling of Archean crustal rock compositions undergoing thermal excursions associated with the Late Heavy Jack Hills Bombardment predicts equally high zircon crystallization temperatures. The lack of such thermal signatures in the Hadean zircon record implies that impacts were not a dominant mechanism of producing the preserved Hadean detrital zircon record. © 2012 Elsevier B.V. All rights reserved.

1. Introduction should persist over time-scales sufficient for crystallization of zircon, provided target rock composition and Zr-content are conducive to Extraterrestrial impacts are thought to have led to lunar formation (ca. zircon formation, with dimensions of >10 μm(Harrison and Watson, 4.5 Ga; Canup and Asphaug, 2001), substantially resurfaced inner solar 1983; Watson, 1996) permitting geochemical and geochronological system bodies at ca. 3.85–3.95 Ga (i.e., the LateHeavyBombardment analysis. Many well-preserved terrestrial impact sites have now been ‘LHB’; Tera et al., 1974), and profoundly influenced the habitability of dated using U–Pb geochronology of impact produced zircon (e.g., Hart early Earth (e.g., Grieve et al., 2006). However, their role in early Earth et al., 1997; Kamo et al., 1996; Krogh et al., 1984; Moser, 1997), but petrogenesis remains poorly understood. Although rare on Earth, due to using zircon to constrain the thermal (Ferry and Watson, 2007; constant resurfacing, impact craters preserved on other terrestrial planets Gibson et al., 1997; Watson and Harrison, 2005) or compositional (Mercury, Mars) and the Moon yield crater size–frequency distributions (Grimes et al., 2007; Maas and McCulloch, 1991)evolutionofimpact suggesting similar early impact histories in the inner solar system (the melt sheet magmas is in its infancy. Such information could provide a Venusian impact record is partially obscured by recent volcanism; baseline for understanding conditions on early Earth and permit com- Neukum et al., 2001). Because of the much greater gravitational cross- parison with the Hadean zircon record (see review in Harrison, 2009) section of the Earth relative to the Moon, it is expected to receive ca. to assess the hypothesis that Hadean detrital zircon formed in impact 20 times higher impact flux (e.g., Grieve et al., 2006). environments. The search for ancient terrestrial impacts has largely focused on shock features in minerals or impact melt spherules (e.g., Cavosie et 2. Background al., 2010; Lowe et al., 2003). We are, however, unaware of any report of such features in Hadean or Archean detritus. Although low energy The long-standing, popular conception that the near-surface Hadean impacts do not produce thick melt sheets, those that result in wide- Earth was an uninhabitable, hellish world (Kaula, 1979; Solomon, 1980; spread decompression melting of the middle crust (Grieve et al., 2006) Wetherill, 1980) has been challenged by the discovery (Compston and Pidgeon, 1986; Froude et al., 1983) and geochemical characterization (see review in Harrison, 2009) of >4 Ga zircon grains from Mt. Narryer and Jack Hills, Western Australia. Hadean zircon crystals may preserve environmental information regarding their formation and thus pro- ⁎ Corresponding author. vide a rare window into conditions on early Earth. Isotopic and

0012-821X/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2012.01.009 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 21 petrologic analyses of these ancient grains have been interpreted to 4. Zircon from preserved terrestrial impactites suggest that early Earth was more habitable than previously envi- sioned, with water oceans, continental crust, and possibly even plate 4.1. Vredefort, South Africa tectonics (Harrison, 2009; Hopkins et al., 2010). Alternatively, diamonds apparently included in these ancient The Vredefort impact structure in South Africa is the largest known zircon grains have been suggested as evidence of their formation terrestrial impact site (Earth Impact Database, 2011) and represents under ultra-high pressure conditions (Menneken et al., 2007; Williams, the deeply eroded remnant of an originally ~300 km wide crater 2007). Diamonds (usually sub-μm) have been found within rocks associ- (Reimold and Gibson, 1996). Although the melt sheet has been eroded ated with terrestrial impact events (Masaitis, 1998)andtheassumed due to post-impact uplift, pseudotachylite breccia and granophyre high bolide flux during the Hadean (Koeberl, 2006) might warrant veins remain. U–Pb dating (isotope dilution thermal ionization mass investigation into the possibility of an impact origin for the Hadean spectrometry ID-TIMS) of zircon isolated from a 45 cm pseudotachylite zircon population if the existence of diamond inclusions can be breccia vein at ~140 m depth within a borehole near the center of the substantiated. remnant crater yield an impact age of 2023±4 Ma (Kamo et al., In this paper, we present U–Pb zircon geochronology, Ti-in-zircon 1996). Similar analysis of zircon extracted from a syn- to post-impact thermometry, and trace-element geochemistry from four of the best- norite dike also yield an age of 2019±2 Ma (Moser, 1997). The target preserved terrestrial impactites to assess the role of impacts in the rocks of the Vredefort impact were the 2.7–3.6 Ga Kaapvaal craton formation of the Hadean detrital zircon. These data reveal a limited granite–greenstone terrane (Schmitz et al., 2004) and sediments and range of formation conditions that strongly contrast with that docu- volcanic rocks of the Witwatersrand basin. mented for Hadean detrital zircon. To further generalize these empir- We obtained zircon and whole rock samples of Vredefort impac- ical results which are necessarily limited by the scarcity of preserved tite from three different sources: (1) pseudotachylite breccia col- impactites, we developed a thermochemical model to obtain a first lected near the quarry at Leeukop, just west of Parys was used for order estimate of the abundance and temperature spectrum of impact mineral separation (VD_PB) as well as in-situ thin section analysis produced zircon that could be expected from a detrital zircon popula- (VD_PB_1A and VD_PB_1B; Fig. 1); (2) granophyre (VD_G; Fig. 1) tion following an LHB-era impact episode on different crustal compo- from the Kommandonek Nature Preserve (provided by Roger Gibson, sitions (Archean vs. modern crust). University of Witwatersrand); and (3) zircon from the INL borehole (VD_INL; Fig. 1) collected just south of the geographical center of the Vredefort Dome (provided by Sandra Kamo, University of Toronto). 3. Methods Pseudotachylite breccia sample VD_PB consists of 1–3cmclasts of Archean granite to granodioritic gneiss target material within a Analysis of zircon was accomplished both in thin section and as fine grained crystalline matrix (Reimold and Gibson, 2006). The sep- separated grains. For all hand samples thin-sections were obtained arated zircon crystals (~35 grains isolated) range in length from 20 and examined via electron imaging using a LEO 1430 VP scanning to 100 μm and exhibit igneous textures (i.e., elongate faceted faces, electron microscope (SEM) to assess zircon crystal size and abun- high axial ratio, pyramidal terminations; Corfu et al., 2003)and dance. Mineral separates were obtained from bulk rock samples by show no planar deformation features. Thin sections (VD_PB_1A and standard heavy liquid separation procedures. Separated zircon crys- VD_PB_1B) were also imaged via SEM to identify zircon (~8 grains) tals were handpicked and mounted in 1 in. diameter epoxy mounts within the melt fraction in-situ and avoid inherited grains associated together with AS3 zircon standard (Paces and Miller, 1993). with target rock clasts. Ion microprobe analyses were conducted with a CAMECA ims1270 Granophyre sample VD_G consists of fine-grained crystalline quartz, using an ~8–12 nA mass-filtered 16O− beam focused to spots between plagioclase and alkali feldspar with long laths of hypersthene and ~20 and 35 μm. We initially screened zircon grains for U–Pb dating to small grains of magnetite (Reimold and Gibson, 2006). Zircon grains discriminate between inherited zircon (i.e., those older than the (~20 grains isolated) average ~100 μm and are fractured. Crystals accepted impact age) and those formed within the impact melt. lack igneous oscillatory zoning and are generally intimately inter- Zircon crystals with U–Pb ages similar to those of previously pub- grown with a Mg-rich pyroxene phase identified by energy disper- lished impact dates (i.e., impact produced zircon) were then re- sive X-ray analysis. analyzed for trace elements (all rare earth elements REE, Ti, and INL borehole (VD_INL) sample was taken at 139.10 m depth, a Hf) within the same area as the U–Pb analysis, ensuring that the fewcmbelowthetopcontactofa~45cmwideintersectionofpseu- data acquired is from the same crystal domain. The trace element dotachylitic breccia (Kamo et al., 1996). We obtained a zircon min- protocol also included mass/charge stations at 26Mg, 55Mn, and 57Fe eral separate of two optically distinct populations of zircon crystals permitting monitoring of beam overlap onto inclusions in zircon that (13 grains total): larger, dark, sub-rounded grains which appear to could overwhelm the low abundances of some critical trace elements have younger overgrowths on old shocked grains, and smaller, clear, in zircon (e.g., Ti). Conditions for trace element analyses were broadly multi-faceted grains that appear to represent new impact produced similar to those of U–Pb dating, with a ~10–15 nA mass-filtered 16O− zircon. beam focused to a ~25–35 μm spot. Analysis yielding high Ti and Fe were re-imaged by SEM to make sure no visible cracks were within 4.2. Sudbury impact, Canada the spot. When cracks were identified grains were re-analyzed on fresh clean surfaces to avoid contamination of Ti associated with crystal The Sudbury Igneous Complex (SIC; ~250 km; Earth Impact Database, defects (Harrison and Schmitt, 2007). NIST 610 glass was used as a 2011)isa2.5–3.0 km thick, elliptical igneous body with four major primary REE standard (Pearce et al., 1997), checked against 91500 subunits from surface to base: granophyre, quartz gabbro, norite, zircon (Harrison and Schmitt, 2007; Wiedenbeck et al., 2004). We contact sublayer (Therriault et al., 2002). U–Pb dating of zircon from used SL13 (6.32±0.3 ppm Ti) and AS3 (21.6±1.6 ppm Ti) as tita- the norite member yield an impact age of 1849.5±0.2 Ma (TIMS; nium standards (Aikman, 2007). For reconnaissance analysis, we Davis, 2008). Target rocks are a mix of Archean granite–greenstone also used a rapid analysis protocol for 48Ti+ and SiO+ (excluding terrains (~2.7 Ga; Krogh et al., 1996) with a small component of other trace elements) by multi-collection with dual electron supracrustal Huronian rocks (Grieve, 1991). multipliers which yield equivalent results. The U–Th–Pb, Ti, and Zircon mineral separates (provided by Don Davis) were obtained REE data are individually tabulated in the Supplementary online from both the mafic (SUD_ M) and felsic norite (SUD_ F) members materials. (Fig. 1) along a transect of the southern exposed limb of the melt 22 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31

VD_PB r e VD_G Par ys iv R a l m a g p a r h V an

Vredefort 27°S VD_INL

R. ruit tsp Rie 20 km

27°30′E

Collar: supracrustalrocks Sample site Core: granitoids B Drill core INL Granophyre A Amphibolite-granulite transition

B D

A C

Sample location

C D

Fig. 1. Sample location for preserved terrestrial impactites. (A) Vredefort Impact, South Africa (after Gibson and Reimold, 2008) showing locations of pseudotachylite breccia and granophyre hand sample (VD_PB, VD_G) as well as drill core INL (VD_INL). (B) Map of Sudbury Impact, Canada and cross-section of the Sudbury Igneous Complex showing norite layer (Sud_F and Sud_M; after Zieg and Marsh, 2005). (C) Morokweng Impact, South Africa (after Jourdan et al., 2010) showing location and cross section of drill core M3 (MK_157 and MK_399). (D) Manicouagan Impact, Canada showing location of main melt sheet sample (MC_MM; after O'Connel-Cooper and Spray, 2011).

sheet (Lightfoot and Zotov, 2005). The mafic and felsic norites are the accepted impact age. Basement target rocks for the Morokweng sheets at the bottom of the 2.5–3.0 km thick SIC, directly above brec- impact structure are primarily Archean granitoids (~2.9–3.0 Ga; Poujol ciated target rocks (Therriault et al., 2002) and consist of hyper- et al., 2002) and meta-volcanics (ca. 2.7 Ga Ventersdorp Lava; Koberl sthene and augite (~2:1), plagioclase, quartz, biotite, and Fe–Ti and Reimold, 2003). oxides (Darling et al., 2009). Separated zircon grains range from Drill core splits and crushed samples (provided by Marco Andreoli) ~20 to >100 μm in length and show igneous textures, with those were obtained from the M3 borehole drilled into the center of the from the felsic norite being more euhedral. Some grains isolated aeromagnetic anomaly (Fig. 1; Jourdan et al., 2010). This drill hole from the mafic norite contain dark domains in SEM imaging that intersects ~870 m of impact melt sheet consisting of differentiated are characterized by high alkali and Fe contents and may reflect granophyric to noritic rocks which overlay brecciated and shocked post-crystallization alteration and were avoided during analysis. basement gneisses (Hart et al., 2002). Core splits from 155, 400, and 1069 m as well as crushed samples associated with the drilling 4.3. Morokweng impact, South Africa processes from 156.5 to 157.98, 341.79–343.29, and 399.20 m were provided for this study. Zircon grains were separated from both the Morokweng, South Africa is a ~70 to 130-km-sized crater (Andreoli 156.5–157.98 and the 399.20 m crushed samples as well as the et al., 2008; Earth Impact Database, 2011). Zircon and biotite from the 1069 m target rock core split (MK_157; MK_399; MK_1069). Sepa- quartz norite melt sheet yield a concordant age of 145±0.8 Ma (U–Pb rated zircon crystals from both crushed samples (~35 grains from TIMS zircon age, 40Ar/39Ar biotite age; Hart et al., 1997), which is each of the 156.5–157.98 and 399.20 m) range in length from 20 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 23 to 80 μm while those separated from the 1069 m target rock core 5.1.2. Sudbury, Canada split tend to be larger, ranging from 50 to 120 μm, with both popu- Ion microprobe analysis of zircon separates from the felsic and lations exhibiting igneous textures (Corfu et al., 2003). mafic norite members of the SIC yield concordant U–Pb ages of 1848±6 Ma (2σ, MSWD=1.7, n=10; Fig. 2), in close agreement 207 206 4.4. Manicouagan impact, Canada with the published Pb/ Pb age of 1849.5±0.2 Ma (Davis, 2008). Only one grain shows evidence for Pb-loss whereby post-analysis The Manicouagan impact, Quebec, is a ~100 km impact crater SEM imaging revealed likely post-crystalline alteration associated (Earth Impact Database, 2011). Post-impact uplift resulted in expo- with high alkali and Fe contents for this crystal. There is no evidence sure of a ~55-km-long island surrounded by an annular, ~5-km-wide within the melt sheet of zircon associated with the Archean granitic lake. Zircon isolated from the upper portions of the melt sheet near gneisses which comprise the target lithologies, and the uniform age the center of the uplift yield an age of 214±1 Ma (TIMS; Hodych and distribution suggests that the impact energy associated with the fi Dunning, 1992). The target rocks for the Manicouagan impact consist Sudbury event was suf cient to fully resorb all inherited zircon. predominately of the Grenville province of the Canadian Shield ranging Consequently, we selected all grains for further analysis. in age from 1.0 to 1.7 Ga (Cox et al., 1998). Hand samples of Manicouagan from the main impact melt sheet 5.1.3. Morokweng, South Africa (MC_MM) and a cross cutting dike from near the base of the melt Zircon grains separated from both the 156.5–157.98 m and the sheet on the west side of Memory Bay (provided by John Spray) are 399.20 m crushed samples yield concordant U–Pb ages of 150±2 Ma homogenous, medium-grained quartz monzodiorite showing only (2σ, MSWD=2.9, n=56; Fig. 2) slightly older than previously reported minor variations in major elements (O'Connel-Cooper and Spray, 2011) ages of 145±0.8 Ma and 144±4 Ma (207Pb/206Pb ID-TIMS zircon age (Fig. 1). Zircon crystals isolated (~40 grains) from the main melt sheet and 40Ar/39Ar biotite age respectively; Hart et al., 1997). We note that (to avoid inherited and altered grains possibly associated with previous U–Pb analysis was performed on zircon from a recrystallized cross cutting dikes) are 40–120 μm in length with prismatic elonga- portion of the quartz norite melt sheet. U–Pb analysis from zircon tions and pyramidal terminations indicative of an igneous origin separated from the 1069 m target rock core split yield concordant (Corfu et al., 2003). ages ranging from ~2.8 to 3.3 Ga, similar to but slightly older than those estimated for the target rocks (~2.7–3.0 Ga; Koberl and Reimold, 2003; Poujol et al., 2002), with half of these grains show- 4.5. Popigai impact, Russia ing evidence for Pb-loss possibly associated with impact shock. Only zircon grains extracted from the 156.5–157.98 m and the The Popigai impact, Siberia, Russia, has an estimated crater diameter 399.20 m depth intervals were selected for further analysis within of ~100 km (Earth Impact Database, 2011). An 40Ar–39Ar age of taga- this study. mite (i.e. impact melt rock) suggests an impact age to 35.7±0.2 Ma (Bottomley et al., 1997) however no zircon grains have been reported with this age. Zircon separated from two hand samples of main melt 5.1.4. Manicouagan, Canada and shocked gneiss (PG_RR~30 grains, PG_SG~70 grains) were sub- U–Pb analysis of zircon crystals separated from the main melt sheet rounded, ovoid to multifaceted grains and range in size from 50 to of the Manicouagan impact yield concordant U–Pb ages of 214±2 Ma 100 μm. (2σ, MSWD=1.7, n=32; Fig. 2) in close agreement with previous studies (214±1 Ma, 207Pb/206Pb TIMS zircon age; Hodych and Dunning, 1992). Zircon crystals show no evidence of Pb loss and inherited 5. Results and discussion grains were absent, suggesting similar conditions to that of Sudbury where the impact melt had sufficient time and heat to fully dissolve 5.1. U–Pb geochronology all relic zircon. All the grains isolated were selected for further analysis. 5.1.1. Vredefort, South Africa Separated zircon crystals from the pseudotachylite breccia sample 5.1.5. Popigai, Russia (VD_PB) yield 207Pb/206Pb ages of ~2.9–3.0 Ga (n=31), with several Zircon separated from both the main melt sheet and the shocked grains (n=6) showing significant Pb-loss, presumably representing gneiss and from the Popigai impact mostly yield Proterozoic concor- – – the Archean target rocks and not neo-crystalline zircon from the dant U Pb ages (~1.8 2.0 Ga) with one grain yielding an Archean impact itself (Reimold and Gibson, 2006; Fig. 2). Eight zircon grains age of ~2.5 Ga (Fig. 2). The majority of zircon ages (~80%) are con- were analyzed in-situ (VD_1A and VD_1B; Fig. 2) within the fine- cordant, with a minor population (~20%) of discordant ages. None grained melt matrix associated with the pseudotachylite breccia of the ~120 crystals analyzed correspond to the 35.7±0.2 Ma age asso- 40 39 to avoid grains from target rock clasts. However, all grains also ciated with the impact event ( Ar/ Ar Tagamite age; Bottomley et al., yield 207Pb/206Pb ages of ~2.8–3.0 Ga, indicative of target rock ages. 1997). This result is consistent with the presumably short cooling duration of such thin (cm to m) veins of pseudotachylite breccia, which are 5.2. Ti-in-zircon thermometry insufficient for resorption of inherited zircon and hence never reach zircon saturation. Similarly zircon crystals isolated from the granophyre Although calculation of accurate magmatic zircon crystallization 207 206 xtln hand sample (VD_G) yield Pb/ Pb ages ranging from ~2.6 to 3.1 Ga temperatures (Tzir ) require knowledge of aTiO2 and aSiO2 activities (n=17; Fig. 2) with less evidence for significant Pb-loss, however still (Ferry and Watson, 2007), we instead ultimately wish to compare representative of the target material and not impact produced zircon. calculated temperatures to detrital Hadean zircon, a population for Although no grains analyzed from the Vredefort samples provide which these parameters are in most cases unknown (Watson and equivalent ages to those published for the impact, two zircon grains Harrison, 2005). Although aTiO2 and aSiO2 activities are largely subu- isolated from the borehole INL yield concordant 207Pb/206Pb ages nity for the samples at hand, equal sub-unity values are directly com- (1978±17 Ma, 2σ, MSWD=0.3, n=2), about 2% lower than TIMS pensatory and magmatic activities of either being ≤0.5 are relatively U–Pb zircon ages of 2020±3 Ma and 2019±2 (Kamo et al., 1996; rare (Ferry and Watson, 2007) and would result in 50–80 °C increases Moser, 1997). We provisionally ascribe this difference to minor Pb in calculated temperatures. Where bulk rock compositions are known, xtln sat loss. Only these two grains were selected for further analysis within Tzir can be compared to zircon saturation model temperatures (Tzir ; this study. Harrison et al., 2007; Watson and Harrison, 1983). 24 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31

0.7

A B 1940 0.35

0.6 3000 1900 1860

U U 0.33 238 238 1820 0.5 2600

Pb / Pb / 1780

206 206 2200 Vredefort 0.31 1740 Sudbury 0.4 1978±17 Ma (MSWD = 0.3) 1848±6 Ma (MSWD = 1.7) 1700 n = 2 n = 12

0.3 0.29 5101520254.24.65.05.45.8 207Pb / 235U 207Pb / 235U

250 0.034 C 0.039 D

200 0.037 0.030 230

U U 0.035

238 0.026 238 160

Pb / 210

Pb / 0.033

206 0.022 206 0.031 Manicouagan Morokweng 190 214±2 Ma (MSWD = 1.7) 120 150±2 Ma (MSWD = 2.9) 0.018 n = 32 n = 56 0.029

0.014 0.027 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.14 0.18 0.22 0.26 0.30 0.34 207Pb / 235U 207Pb / 235U 0.6 E

2600

0.4 2200 1800

238 1400 Popigai

Pb / 0.2 1000 No impact-produced zircon 206 n = 102

600

0.0 0481216 207Pb / 235U

Fig. 2. Concordia diagrams of zircon extracted from preserved terrestrial impactites calibrated against AS3 zircon standard (Paces and Miller, 1993). (A) Nearly all grains extracted from Vredefort are inherited from the target with exception of two grains from the INL borehole (denoted in red) which yield ages similar to the published impact age (~2020 Ma; 207Pb/206Pb TIMS zircon age; Kamo et al., 1996; Moser, 1997). (B) Zircon from the maﬁc norite of the Sudbury Igneous Complex yield ages in agreement with the published impact age (1849.5±0.2 Ma; 207Pb/206Pb TE-TIMS zircon age; Davis, 2008) with no signs of inheritance from the target rock. (C) Grains from the Morokweng borehole M3 yield ages similar to, however slightly older than, the published impact age (145±0.8 Ma; 207Pb/206Pb ID-TIMS zircon age, 40Ar/39Ar biotite age; Hart et al., 1997)withno sign of inheritance. (D) Zircon from the Manicouagan impact yield ages identical to those from previous studies (214±1 Ma; 207Pb/206Pb TIMS zircon age; Hodych and Dunning, 1992) and show no inheritance also. (E) No impact produced zircon has been extracted from the Popigai samples provided for this study with all grains reﬂecting target rock ages.

5.2.1. Vredefort, South Africa analysis. Ti concentrations results for these two grains are 11±0.7 xtln Of the over than 60 grains analyzed from three separate samples and 20±1.1 ppm which corresponds to Tzir of 750±15 and 810± of Vredefort, only two grains isolated from the INL drill core provided 15 °C, respectively (Watson and Harrison, 2005). All other grains ages suitable (i.e. within 10% of the published impact age) for further dated from Vredefort samples are presumed to be associated with M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 25

Fig. 3. Titanium concentrations and Ti-in-zircon crystallization temperature, assuming αTiO2 =1 (Watson and Harrison, 2005), of impact produced zircon. Saturation temperature underestimates crystallization temperature for differentiated impact melts (i.e. Sudbury and Morokweng; Watson and Harrison, 2005) (A) Sudbury impact produced zircon xtln sat xtln sat Tzir =748 ± 60 °C, bulk norite Tzir =650 °C (calculated from Lightfoot and Zotov, 2005). (B) Morokweng impact produced zircon Tzir =797±70 °C, bulk melt sheet Tzir =730°C xtln sat (calculated from Andreoli et al., 1999). (C) Manicouagan impact produced zircon Tzir =747±35 °C, bulk melt sheet Tzir =752°C(calculatedfrom Floran et al., 1978). 26 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31

sat target rock zircon and do not ﬁt the criteria for an impact produced et al., 1978) which predict Tzir ~752 °C. This agreement is consistent zircon. Although these results agree well with the bulk zircon satura- with Manicouagan being an undifferentiated impact melt. Whereas tion temperature of 772 °C calculated from Reimold (1991) for the recent studies suggest possible pockets of differentiation throughout pseudotachylite breccia, more impact formed zircon grains are the melt (O'Connel-Cooper and Spray, 2011), we did not observe xtln sat needed to verify this result. This result (Tzir ~ Tzir ) is also consistent this in the sampled zircon population. In those cases where Hf is with the lack of differentiation expected within such small veins of reported but Zr is not (e.g., Simonds et al., 1978), Zr content is esti- melt. mated using an average Zr/Hf=37 (Rudnick and Gao, 2003).

5.2.2. Sudbury, Canada 5.3. Trace elements Zircon separated from the felsic (n=12) and mafic norite (n=13) members of the SIC yield overlapping average Ti concentrations of Results from REE analysis of all impactite-produced zircon are, as xtln 10 ±4 and 12 ±7 ppm, corresponding to Tzir of ~740 and ~760 °C, expected, similar to terrestrial igneous zircon from a broad range of respectively (Fig. 3a). There is relatively little within sample varia- ages (Quaternary to Hadean; Grimes et al., 2007; Peck et al., 2001; tion with >75% of the data plotting within 700–780 °C. This temper- Fig. 4). They are characterized by a relatively steep slope from LREE ature is slightly lower than that given in an earlier study of Sudbury to HREE, and display positive Ce- and negative Eu-anomalies. The (~800 °C for zircon grains isolated from the norite member and ana- positive Ce-anomaly is commonly interpreted as evidence for oxi- lyzed LA–ICPMS; Darling et al., 2009). This relatively small difference dized magmas in which partitioning of Ce4+ dominates over Ce3+ could potentially be due to Ti hosted in cracks (Harrison and Schmitt, (Hoskin and Schaltegger, 2003). The negative Eu-anomaly almost 2007) which are more likely to be encountered in the larger analysis certainly reflects plagioclase fractionation of the parent magma as xtln volume required by laser ablation as opposed to ion sputtering. Tzir Eu easily substitutes for Ca within plagioclase (Hoskin and Schaltegger, sat agrees well with the calculated Tzir of ~650 °C for bulk norite composi- 2003), and plagioclase is ubiquitously present in large impact target tions (Lightfoot and Zotov, 2005) in that such temperatures are typically rocks. Differences in REE contents of differing impact sites likely reflect ~50–100 °C lower than that expected for the onset of zircon crystalliza- local concentration variations. While they are consistent within target tion in a fractionating magma (Harrison et al., 2007), due to the much populations, we emphasize that REE signatures of zircon cannot be lower M, where M = (2Ca+Na+K)/(Al·Si), present in the melt during used to clearly identify zircon provenance. zircon crystallization as opposed to that calculated from the bulk rock The trace element compositions of impact-produced zircon from composition. Darling et al. (2009) reported elevated REE contents within Sudbury, Morokweng and Manicouagan (Fig. 5) fall generally within lower temperature zircon, also suggesting a fractional crystallization the fields (in U vs. Yb, U/Yb vs. Hf or Y space) defined by zircon crys- process causing REE enrichment within the residual melt; however an tallized from continental crust as do detrital Hadean zircon from the anti-correlation between Ti abundance and REE is absent in our data, Jack Hills region (Grimes et al., 2007). Trace elements may also be possibly due to the restriction of our sample to the basal norite layer. useful in discriminating between neoformed and reset zircon (Abramov et al., 2011). For example, if elements that diffuse slower than Pb (e.g., 5.2.3. Morokweng, South Africa Ti and REE; Cherniak and Watson, 2003; Cherniak et al., 1997)are Average Ti concentrations of zircon crystals separated from 157 to distinctly different between melt sheet zircon and those from the 399 m within borehole M3 are 25±19 and 15±9 ppm respectively target rock, it would appear likely that the zircon crystals are neo- (n=31 for 157 m sample; n=19 for 399 m sample). This corre- formed. REE patterns in zircon analyzed from the Morokweng M3 xtln sponds to a Tzir of ~830 °C for the 157 m sample and ~780 °C for drill core target rock material (1069 m) show a clear distinction (pri- the 399 m sample and an overall average of 797±70 °C (Fig. 3b). Rel- marily in the lack of a HREE enrichment) from those isolated from xtln atively little variation in Tzir is observed from the 399 m sample, the melt sheet, reflecting the different geochemical conditions dur- although significant variations exist within the 155 m sample with ing crystallization of impact produced zircon and target rock zircon. xtln Tzir ranging from 720 to 990 °C. High sensitivity ion imaging (Harrison and Schmitt, 2007) may be helpful in ruling out contam- 5.4. Zircon saturation ination as a cause of the high apparent temperatures (although no cracks are apparent in SEM images).Morokweng is a differentiated The crystallization temperature spectrum of impact produced xtln impact melt sheet (Hart et al., 2002) and thus the zircon saturation zircon from four terrestrial impacts (average Tzir =773±87 °C) is of ~730 °C (Andreoli et al., 1999) underestimates the onset of zircon remarkably consistent with the average zircon saturation tempera- sat crystallization by ~50–100 °C (Harrison et al., 2007). The M3 bore- ture (Tzir ) of ca. 780 °C calculated by Watson and Harrison (2005) hole extends through nearly 870 m of impact melt sheet (Hart et from a large (~19,000) database of whole rock samples across al., 2002) and only samples from the upper half (157 and 399 m) Australasia. Estimates of bulk continental crust composition yield sat are used within this study. Impact melt sheets tend to cool via the similar Tzir values (ca. 780 °C, Rudnick and Gao, 2003;ca.760°C, inward propagation of solidification fronts from both the surface Gao et al., 1998). Since it is broadly assumed that impact melts and the basal layer (Zieg and Marsh, 2005). largely reflect decompression melting of the middle crust (Melosh, 1989), we also note that estimates of the composition of modern sat 5.2.4. Manicouagan, Canada middle crust yield Tzir of ca. 740 °C (Rudnick and Fountain, 1995; Zircon separated from the main melt sheet of Manicouagan Rudnick and Gao, 2003). Although the observed apparent crystalliza- xtln yield an average Ti concentration of ~12±6 ppm, corresponding tion temperatures (Tzir ) for impact produced zircon are consistent xtln sat to a Tzir of 747±35 °C (n=32; Fig. 3c). Crystals analyzed in repli- with Tzir for present crustal compositions, they appear to be distinctly cate using single- and multi-collection techniques yield similar higher than that for detrital Hadean zircon population (i.e., ~680 °C; values, with the exception of four crystals where single-collection Watson and Harrison, 2006). On the assumption that the Hadean Ti data are higher relative to the same grains analyzed in multi- crust was compositionally closer to Archean than modern crust, we sat collection (possibly due to beam overlap onto cracks). Averages determined Tzir for Archean rocks in the GEOROC database with for each method agree well (Ti average concentration=11.4 ppm sufficient compositional data to permit calculation (n=10,059). sat in multi-collection; Ti average concentration =12.2 ppm in single- The resulting Tzir distribution is bimodal with peaks at ca. 600 °C collection). and 775 °C with an average value of 690 °C. sat The Ti-in-zircon temperatures agree well with zircon saturation Although this average Tzir is close to the Hadean value, several in- xtln temperatures calculated from averaged bulk rock analyses (Floran terpretive complexities warrant discussion when comparing Tzir M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 27

5 5 10 10

4 4 10 10

3 3 10 10

2 2 10 10 Manicouagan 1 1 10 Vredefort Main Melt Sheet 10 Sample/chondrite INL Borehole (n= 30) 0 0 10 (n=2) 10

-1 -1 10 10 5 5 10 10

4 4 10 10

3 3 10 10

2 2 10 10

1 1 10 Sudbury 10

Sample/chondrite Norite Impact Layer Morokweng 0 M3 Borehole 0 10 (n=12) 10 Main Melt Sheet (n=41)

-1 Target (n=11) -1 10 10 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Element Element

Fig. 4. Chondrite-normalized rare earth element abundances in impact produced zircon from ion microprobe measurements. REE patterns are indistinguishable from igneous zircon of all ages (Quaternary to Hadean; Hoskin and Schaltegger, 2003; Peck et al., 2001; Grimes et al., 2007), characterized by a relatively steep slope from LREE to HREE and displaying a positive Ce-anomaly and a negative Eu-anomaly.

sat xtln and Tzir spectra. We first note that Tzir are sensitive to the activities To calculate the expected abundance of impact produced zircon of silica and rutile during zircon crystallization (i.e., an apparent sub- within a detrital LHB-era zircon population we have used results of minimum melting temperatures could reflect either sub-unity activ- the thermal model of Abramov and Mojzsis (2009). They constructed xtln ities or sub-solidus growth). Second, the expected Tzir for zircon a numerical framework to assess the amount of crustal heating result- crystallizing from fractionating magmas are expected to be 50– ing from an LHB impact history. Although their goal was to assess 100 °C higher than that predicted from zircon saturation calculations variations in the scale of the subsurface micro-biosphere due to the (Harrison et al., 2007). These complications notwithstanding, the LHB, we can use their data to estimate crustal temperature changes sat correspondence between average crustal Tzir and the impact melt that would lead to melting and re-precipitation of zircon. formed zircon reported here suggests that the small population of Using the LHB impact flux inferred from the lunar cratering record preserved terrestrial impact melts may be representative of a global and asteroid belt size distribution, Abramov and Mojzsis (2009) average. found that less than 10% of the lithosphere would have experienced The impact-melt formed zircon temperature distribution partially temperature increases (ΔT)of≥500 °C. However, under an elevated overlaps the Gaussian-form Hadean temperature distribution but the Archean geotherm, this could permit a significant fraction of the crust overall misfit does not support an impact origin for the Hadean zircon to achieve anatexis. Assuming a surface temperature of 0 °C and grains. A Kolmogorov–Smirnov test yields p=0.002, indicating that Archean geothermal structure that reaches 800 °C at 40 km (e.g., the two distributions are distinct populations at a high level of confi- Condie, 1984), we coupled the thermal anomaly data of Abramov dence (Fig. 6). and Mojzsis (2009) with those samples from the Archean GEOROC database containing sufficient data to calculate zircon saturation xtln 5.5. Modeling Tzir of impact melt formed zircon temperatures to create a stochastic model of zircon formation temperatures during the LHB. In our model we ignored the limited po- The LHB is generally interpreted as a relatively brief spike in tential of oceanic crust to preserve impact formed zircon, implicitly impact flux (Gomes et al., 2005), although it remains possible that assume that the Archean GEOROC database is representative of it instead reflects the terminal phase of a protracted cataclysm continental crust between 4.0 and 3.8 Ga, and limit the upper tem- (Hartmann, 1975). In either interpretation, the terrestrial impact perature in the ΔT distribution to 1200 °C. The latter assumption is flux between ca. 4.0 and 3.8 Ga would be broadly similar. Thus justified on the basis that this exceeds all predicted solidus tem- givenanestimateofthatflux, we should be able to estimate the peratures (Tsol) in the database and only 1% of the lithosphere is xtln resulting Tzir spectrum and approximate the fraction of Zr in the predicted to have reached ≥1200 °C (Abramov and Mojzsis, 2009). continental crust that was processed through impact melting. We Zircon solubility in crustal melts is a simple function of zirconium can then use these calculations to estimate the fraction of detrital content [Zr], temperature, and rock chemistry (i.e., M=(2Ca+Na+ zircon from a post-LHB terrane that would bear the thermal signature K)/(Al·Si); Harrison and Watson, 1983; Watson and Harrison, 1983). xtln of an impact origin and to compare with the Hadean Tzir distribution Thus knowledge of M, [Zr], T′ (i.e., ambient temperature plus the ΔT as- as a partial test of whether they could reflect an impact origin. sociated with impacts), and the relationship between M and melting 28 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31

10000 Hadean zircons (Harrison, 2009) Mafic zircons (Fu et al., 2008) Model zircon temperatures (this study) Impact produced zircons (this study) 1000

Continental Field

100 U (ppm) Relative probability

10 Morokweng - M3 Borehole (n=44) Sudbury - Norite Layer (n=12) Manicouagan - Main melt (n=30) 1 600 700 800 900 1 10 100 1000 10000 100000 Temperature (oC) Yb (ppm) 100 Fig. 6. Probability density function for Ti-in-zircon crystallization temperatures of Ha- dean zircon from Western Australia (n=69; Harrison, 2009), maﬁc zircon (n=304; Fu et al., 2008), impact produced zircon (n=111; this study) and modeled impact produced zircon temperatures (n=80; this study). The signiﬁcant temperature contrast between impact produced zircon, modeled impact produced zircon (on an Archean tar- 10 get composition) and that observed for Hadean zircon grains suggests that impacts were not a major contributor to the Hadean population. Continental Field

1 To obtain a relationship between M and melting conditions, we Continental Field ﬁ U/Yb rst note that the GEOROC database of analyzed Archean rocks (re-

stricted to SiO2 values between 35 and 85% to preclude sampling of non-silicates and quartzites) plots in a coherent fashion with a nega-

0.1 tive slope on a M vs. SiO2 diagram (Fig. 7). We take this as evidence Morokweng - M3 Borehole (n=44) that this parameter appropriately characterizes petrogenic variation Sudbury - Norite Layer (n=12) Manicouagan - Main melt (n=30) in this broad suite of rocks. Also shown in Fig. 7 is the regression of M vs. SiO2 of the felsic, intermediate and maﬁc rocks utilized in the 0.01 classic melting studies of Wyllie (1977) which essentially reproduce 5000 10000 15000 20000 25000 30000 35000 the Archean database trend. Using either the 5 kbar T or 5% H O Hf (ppm) sol 2 (i.e., the average water content of Archean GEOROC samples is ~3%) 100 liquidus temperatures (Tliq) from Wyllie (1977) permits us to trans- late M into a melting or crystallization temperature (note that the calculations are not sensitive to assumed P). To obtain impact-associated temperatures, we randomly coupled 10 an ambient temperature (between 0° and 800 °C) with a ΔT value scaled to the distribution given in Fig. 3 of Abramov and Mojzsis Continental Field (2009) (e.g., a ΔT of 50 °C was encountered 40 times more often than 1000 °C). We then arrayed the 10,059 samples in the database 1 and randomly linked each synthetic impact temperature with an

U/Yb analyzed rock. Thus each rock is associated with an M and [Zr], a

calculated solidus (Tsol) and liquidus (Tliq)temperature(i.e.,viathe Mvs.T relationships), and a value of T′.NotethatMwillbemuch 0.1 lower than the bulk rock value for low melt fractions. This will be Morokweng - M3 Borehole (n=44) particularly pronounced for maﬁccompositionsthusrequiringpa- Sudbury - Norite Layer (n=12) rameterization of M with melt fraction (described below based on Manicouagan - Main melt (n=30) synthetic experiments using MELTS; Ghiorso and Sack, 1995). The 0.01 10 100 1000 10000 100000 following logical statements were then applied to each of the col- Y (ppm) lective data sets in the sequence:

Fig. 5. Trace element plots of impact produced zircon from Morokweng, Sudbury and 1) Did the rock melt due to impact (i.e., is T′>Tsol)? If not, the event Manicouagan to discriminate between continental and oceanic target rock composi- does not produce magmatic zircon. If true, T′ is recorded and the tions. All grains are corrected for uranium decay. Continental field from Grimes et al. calculation continues. (2007) with lower boundary denoted by a solid line on each diagram, indicating the sat b upper most limit of unambiguously oceanic zircon. End points for lines: U vs. Yb (25, 1), 2) Is Tzir Tsol? If true, the whole rock composition will not saturate (20,000, 10,000); U/Yb vs. Hf (5000, 0.05), (35,000, 5); U/Yb vs. Y (200, 0.01), (100,000, zircon and the calculation terminates unless T′>Tliq, in which 5). Nearly all impact produced zircon suggest continental target lithologies however case we assume the melt differentiates and thus produces zircon Manicouagan plots nearly in the oceanic field possibly due to a slightly more mafic target. xtln sat with Tzir =Tzir +50°C if >Tsol (i.e., we account for the high- conditions permits any random association of analyzed Archean rock temperature onset of zircon saturation in crystallizing interme- and possible T′ to be evaluated for the potential to produce impact diate to mafic melts by adding the conservative lower limit of xtln sat zircon. the observed 50–100 °C difference between Tzir and Tzir (see M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 29

1100 1050 1000 950 900 850 800 750 and felsic inclusion assemblage (i.e., quartz+muscovite) that characterizes the vast majority of the detrital Hadean zircon population is o 950 900 850 800 750 700 ( C) indicative of their origin at near water saturated granitoids. Given our model result, caution should be exercised when drawing infer- 0 ences regarding whole-rock mineralogy from inclusion assemblages. 4 5 Indeed, Darling et al. (2009) found rare muscovite inclusions in im- 9 xtln 12 pact produced Sudbury zircon crystals. Nonetheless, the high Tzir 15 found by both us and Darling et al. (2009) are distinctively higher 18 20 than the detrital Hadean zircon population to effectively rule out an 3 22 impact source as a signiﬁcant contributor to their origin. 36

6. Conclusions 2 Impacts investigated in this study contain neo-formed (Sudbury, M (Na+K+2Ca)/(AIxSi) Morokweng, Manicouagan, Vredefort) or entirely inherited zircon (Popigai), with no obviously relationship between crater diameter 1 Wyllie (1977) Impact melt sheets (this study) and zircon occurrence. Inherited grains yield ages consistent with local target rocks and their survivability reflects either low target rock [Zr] (i.e., never allowing zircon saturation), rapidly cooling melt sheets 50 60 70 80 (i.e., a kinetic barrier to dissolution), or low energy impacts that never SiO (wt%) 2 completely melted the middle crust. We note that although Vredefort is currently the largest known terrestrial impact site, erosion and resur- Fig. 7. GEOROC Archean database (n=10,058) plotted as M vs SiO2. Top axes are experimental results from Wyllie (1977) at 5 kbar, whereby the upper and lower axes gence of the central dome have removed much of the presumably large represent liquidus temperatures at 5 wt.% H2O and water-undersaturated solidus melt sheet that once existed (Therriault et al., 1997) explaining the fi temperatures, respectively. Wyllie (1977) dashed line is a tofMvs.SiO2 for melting rarity of impact produced zircon in the samples studied. experiments using felsic, intermediate and mafic compositions. This fitcoincideswith U–Pb geochronology of zircon from the Sudbury and Manicouagan the regression of M vs. SiO for bulk composition of impactites from this study, which 2 impact melts agree well with published ages, however those for underscores the validity of using experimental melting relations to model impactite formation. Vredefort and Morokweng do not (no impact produced zircon was observed within the Popigai samples). An impact age of ~1980 Ma for Vredefort is younger than the previously published age of ~2020 Ma, sat sat Harrison et al., 2007)). If Tzir >Tsol and Tsol ≤T′≤Tzir ,thenT′ is taken to probably reflecting minor Pb loss. Morokweng U–Pb zircon ages xtln be Tzir . (~150 Ma) are slightly older than previously published TIMS U–Pb 3) In cases where Tsol bT′bTliq, we scaled M according to the relation- age of ~145 Ma. No relic zircon grains were discovered associated ships M′=0.4 M if Tsol bT′b(Tliq −Tsol)·0.5+Tsol,orM′=0.8 M if with the target rocks ruling out the possibility of inheritance and Tsol bT′b(Tliq −Tsol)·0.5+Tsol (these relationships approximate leaving open the possibility that the published ages reflect post- the compositional response of the Wyllie (1977) rock analyses to impact effects. melting using the MELTS algorithm). Impact-produced zircon is indistinguishable from that of igneous or Hadean grains in REE, however plots of U vs. Yb and U/Yb vs. Hf Only about 8% within each run batch resulted in conditions per- or Y suggest that some inference can be made to the target rock com- mitting zircon formation. Thus in order to attain a robust result, we position (continental vs. oceanic crust). Trace element geochemistry xtln executed the algorithm 1000 times. The average Tzir resulting from can be used to distinguish between grains that have had their ages this calculation is ~783 °C — essentially identical to our observed reset to that of the impact and those that crystallized from the impact xtln value. The modeled Tzir spectrum is shown in Fig. 6.Asurprising melt, assuming that magmatic conditions within the melt sheet are result to us was that the average M value of the zircon forming different than those of the target rocks. xtln events is only 1.1 with no value exceeding 2 (i.e., felsic sources Crystallization temperatures (Tzir ) of 111 zircon crystals separated dominate). This indicates that, although mafic samples (which domi- from Sudbury, Vredefort, Morokweng, and Manicouagan impacts indi- nate the database) can contain substantial [Zr], their high melting cate an average crystallization temperature of 773±87 °C. Based on temperatures and solidi restrict their ability to contribute signifi- the Ti-in-zircon thermometry results, we conclude that zircon derived cantly to impact produced zircon in the crust. Also of note is that from impactites do not represent a dominant source for the Hadean xtln the average Tzir of 780 °C is significantly higher than the average grains thus far documented from Western Australia. sat xtln Tzir of 690 °C. As a test of robustness, we also ran the model using As expected, impact produced zircon Tzir values are consistent the same large database from which Watson and Harrison (2005) with that calculated from bulk rock zircon saturation systematic for sat xtln reported an average Tzir of 780 °C and obtained an average Tzir of the one presumably undifferentiated melts (Manicouagan). Zircon xtln 810 °C. Despite model simplifications, the consistency and relative Tzir values for Sudbury and Morokweng – both differentiated bodies – xtln sat sat order of results from the two databases (i.e., Tzir >Tzir ) suggests are 50–100 °C higher than Tzir , consistent with previous observations. that the model is at the least internally consistent and likely pre- A Monte Carlo model relating impact thermal anomalies associated xtln dicts an accurate temperature spectra. Thus we infer that the zircon with the LHB with Archean rock chemistry supports our Tzir estimate – crystallization temperature spectra of our necessarily limited em- despite its limited size – as being globally representative. The signif- pirical results is broadly representative of that produced globally icant temperature contrast between modeled impact formed zircon in crustal impact melts. on Archean crust and that observed for Hadean zircon grains effec- An implication of this model is that the relative absence of mafic tively rules out an impact source as a significant contribution to targets contributing to the zircon temperature distribution could the Hadean population. The model developed in this study can lead to incorrect inferences regarding the nature of the crust drawn also be applied to lunar compositions to predict the crystallization from the chemistry and inclusion assemblages in detrital zircon. For spectrum associated with an LHB event on the lunar surface and xtln example, Harrison (2009) emphasized that the low Tzir distribution compared to results from lunar zircon (e.g., Taylor et al., 2009). 30 M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31

Supplementary materials related to this article can be found on- Gao, S., Luo, T., Zhang, B., Zhang, H., Han, Y., Zhao, Z., Hu, Y., 1998. Chemical composition of the continental crust as revealed by studies in East China. Geochim. Cosmo- line at doi:10.1016/j.epsl.2012.01.009. chim. Acta 62, 1959–1975. Ghiorso, M.S., Sack, R.O., 1995. Chemical mass transfer in magmatic processes. IV. A revised and internally consistent thermodynamic model for the interpolation Acknowledgments and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contrib. Mineral. Petrol. 119, 197–212. The data published in this study would not have been possible Gibson, R.L., Armstrong, R.A., Reimold, U.W., 1997. The age and thermal evolution of the Vredefort impact structure: a single-grain U–Pb zircon study. Geochim. Cosmo- without the generous help and sample provision from: Donald W. chim. Acta 61, 531–1540. Davis and Sandra L. Kamo (University of Toronto), John G. Spray Gomes, R., Levison, H.F., Tsiganis, K., Morbidelli, A., 2005. Origin of the cataclysmic Late (University of New Brunswick), Doreen Ames (Geological Survey of Heavy Bombardment period of the terrestrial planets. Nature 435, 466–469. Grieve, R.A.F., 1991. The Sudbury Structure: controversial or misunderstood? J. Geo- Canada), Roger A. F. Gibson (University of Witwatersrand), Marco phys. Res. 96, 22,753–22,764. A. G. Andreoli (South African Nuclear Energy Corporation), W. Uwe Grieve,R.A.F.,Cintala,M.J.,Therriault,A.M.,2006.Large-scaleimpactsandtheevolu- Reimold (Berlin Museum of Natural History), Victor Masaitis (Karpinsky tion of the Earth's crust: the early years. In: Reimold, W.U., Gibson, R.L. (Eds.), Processes of the Early Earth: Geological Society of America Special Paper, 405, Russian Geological Research Institute), Boris E. Burakov, (VG Khlopin pp. 22–31. Radium Institute), Xiaonan Lu (Boston College) and the UCLA SIMS Grimes, C.B., John, B.E., Kelemen, P.B., Mazdab, F.K., Wooden, J.L., Cheadle, M.J., Hanghoj, K., group. Support for this study was also provided by the Eugene M. Schwartz, J.J., 2007. Trace element chemistry of zircons from oceanic crust: a method for distinguishing detrital zircon provenance. Geology 35, 643–646. Shoemaker Impact Cratering Award from the Planetary Geology Harrison, T.M., 2009. The Hadean Crust: evidence from >4 Ga zircons. Ann. Rev. Earth Division of the Geological Society of America. The ion microprobe Planet. Sci. 37, 479–505. facility at UCLA is partly supported by a grant from the Instrumen- Harrison, T.M., Schmitt, A.K., 2007. High sensitivity mapping of Ti distributions in – tation and Facilities Program, Division of Earth Sciences, National Hadean zircons. Earth Planet. Sci. Lett. 261, 9 19. Harrison, T.M., Watson, E.B., 1983. Kinetics of zircon dissolution and zirconium dif- Science Foundation. fusion in granitic melts of variable water content. Contrib. Miner. Petrol. 84, 66–72. Harrison, T.M., Watson, E.B., Aikman, A.B., 2007. Temperature spectra of zircon crystal- References lization in plutonic rocks. Geology 35, 635–638. Hart, R.J., Andreoli, M.A.G., Tredoux, M., Moser, D., Ashwal, L.D., Eide, E.A., Webb, S.J., Abramov, O., Mojzsis, S.J., 2009. Microbial habitability of the Hadean Earth during the Brandt, D., 1997. Late Jurassic age for the Morokweng impact structure, southern late heavy bombardment. Nature 459, 419–422. Africa. Earth Planet. Sci. Lett. 147, 25–35. Abramov, O., Kring, D.A., Mojzsis, S.J., 2011. Modeling of impact-induced age resetting Hart, R.J., Cloete, M., McDonald, I., Carlson, R.W., Andreoli, M.A.G., 2002. Siderophile- and partial Pb-loss in zircon grains. 42nd Lunar and Planetary Science Conference, rich inclusions from the Morokweng impact melt sheet, South Africa: possible frag- p. 2674. ments of a chondritic meteorite. Earth Planet. Sci. Lett. 198, 49–62. Aikman, A.B., 2007. Tectonics of the eastern Tethyan Himalaya. Ph.D. thesis. Australian Hartmann, W.K., 1975. Lunar “cataclysm”: a misconception? Icarus 24, 181–187. National University. Hodych, J.P., Dunning, G.R., 1992. Did the Manicouagan impact trigger end-of-Triassic Andreoli, M.A.G., Ashwal, L.D., Hart, R.J., Huizenga, J.M., 1999. Petrology and geo- mass extinction? Geology 20, 51–54. chemistry of nickel and iridium-rich quartz norite from the late Jurassic Morokweng Hopkins, M.D., Harrison, T.M., Manning, C.E., 2010. Constraints on Hadean geody- impact structure, South Africa. In: Dressler, B.O., Sharpton, V.L. (Eds.), Large Meteorite namics from mineral inclusions in >4 Ga zircons. Earth Planet. Sci. Lett. 298, Impacts and Planetary Evolution 11, pp. 91–108. 367–376. Andreoli, M.A.G., Hart, R.J., Cooper, G.R., Webb, S.J., 2008. The Morokweng impact Hoskin, P.W.O., Schaltegger, U., 2003. The composition of zircon and igneous and meta- crater, South Africa: a complex, multiring structure with a ~130 km radius exter- morphic petrogenesis. Rev. Mineral. Geochem. 53, 27–62. nal ring and asymmetric radial sectors. Large Meteorite Impacts and Planetary Jourdan, F., Andreoli, M.A.G., McDonald, I., Maier, W.D., 2010. 40Ar/39Ar thermochronol- Evolution IV. paper id 3048. ogy of the fossil LL6-chondrite from the Morokweng crater, South Africa. Geochim. Bottomley, R., Grieve, R., York, D., Masaitis, V., 1997. The age of the Popigai impact Cosmochim. Acta 74, 1734–1747. event and its relation to events at the Eocene/Oligocene boundary. Nature 388, Kamo, S.L., Reimold, W.U., Krogh, T.E., Colliston, W.P., 1996. A 2.023 Ga age for the 365–368. Vredefort impact event and a first report of shock metamorphosed zircons in Canup, R.M., Asphaug, E., 2001. Origin of the Moon in a giant impact near the end of the pseudotachylitic breccias and Granophyre. Earth Planet. Sci. Lett. 144, 369–388. Earth's formation. Nature 412, 708–712. Kaula, W.M., 1979. Thermal evolution of Earth and Moon growing by planetesimal im- Cavosie, A.J., Quintero, R.R., Radovan, H.A., Moser, D.E., 2010. A record of ancient cata- pacts. J. Geophys. Res. 84, 999–1008. clysm in modern sand: shock microstructures in detrital minerals from the Vaal Koberl, C., Reimold, W.U., 2003. Geochemistry and petrography of impact breccias and River, Vredefort Dome, South Africa. GSA Bull. 122, 1960–1980. target rocks from the 145 Ma Morokweng impact structure, South Africa. Geo- Cherniak, D.J., Watson, E.B., 2003. Diffusion in zircon. Rev. Mineral. Geochem. 53, chem. Cosmochem. Acta 67, 1837–1862. 113–143. Koeberl, C., 2006. Impact processes on the early Earth. Elements 2, 211–216. Cherniak, D.J., Hanchar, J.M., Watson, E.B., 1997. Rare-earth diffusion in zircon. Chem. Krogh, T.E., Davis, D.W., Corfu, F., 1984. Precise U–Pb zircon and baddeleyite ages for Geol. 134, 289–301. the Sudbury area. In: Pye, E.G., Naldrett, A.J., Giblin, P.E. (Eds.), The geology and Compston, W., Pidgeon, R.T., 1986. Jack Hills, evidence of more very old detrital zircons ore deposits of the Sudbury structure: Ontario Geologic Survey Special, Volume in Western Australia. Nature 321, 766–769. 1, pp. 431–446. Condie, K.C., 1984. Archean geotherms and supracrustal assemblages. Tectonophysics Krogh, T.E., Kamo, S.L., Bohor, B.F., 1996. Shock metamorphosed zircons with 105, 29–41. correlated U–Pb discordance and melt rocks with concordant protolith ages indi- Corfu,F.,Hanchar,J.M.,Hoskin,P.W.O.,Kinny,P.,2003.Atlasofzircontextures.In: cateanimpactoriginfortheSudburyStructure.In:Basu,A.,Hart,S.(Eds.), Hanchar, J.M., Hoskin, P.W.O. (Eds.), Zircon. : Reviews in Mineralogy and Geo- Earth Processes — Reading the Isotopic Code: Geophysical Monograph, 95, pp. chemistry, 53. Mineralogical Society of America, pp. 470–500. 343–354. Cox, R.A., Dunning, G.R., Indares, A., 1998. Petrology and U–Pb geochronology of mafic, Lightfoot, P.C., Zotov, I.A., 2005. Geology and geochemistry of the Sudbury Igneous high-pressure, metamorphic coronites from the Tshenukutish domain, eastern Complex, Ontario, Canada: origin of nickel sulfide mineralization associated with Grenville Province. Precambrian Res. 90, 59–83. an impact-generated melt sheet. Geol. Ore Deposits 47, 349–381. Darling, J., Storey, C., Hawkesworth, C., 2009. Impact melt sheet zircons and their impli- Lowe, D.R., Byerly, G.R., Kyte, F.T., Shukolyukov, A., Asaro, F., Krull, A., 2003. Spherule cations for the Hadean crust. Geology 37, 927–930. Beds 3.47–3.24 billion years old in the Barberton Greenstone Belt, South Africa: a Davis, D.W., 2008. Sub-million-year age resolution of Precambrian igneous events by record of large meteorite impacts and their influence on early crustal and biologi- thermal extraction–thermal ionization mass spectrometer Pb dating of zircon: ap- cal evolution. Astrobiology 3, 7–48. doi:10.1089/153110703321632408. plication to crystallization of the Sudbury impact melt sheet. Geology 36, 383–386. Maas, R., McCulloch, M.T., 1991. The provenance of Archean clastic metasediments Earth Impact Database, http://www.passc.net/EarthImpactDatabase/index.html,accessed in the Narryer gneiss complex, Western Australia: trace element geochemistry, July 2011. Nd isotopes, and U–Pb ages from detrital zircons. Geochim. Cosmochim. Acta Ferry, J.M., Watson, E.B., 2007. New thermodynamic models and revised calibrations 55, 1915–1932. for the Ti-in-zircon and Zr-in-rutile thermometers. Contrib. Mineral. Petrol. 154, Masaitis, V.L., 1998. Popigai crater: origin and distribution of diamond-bearing impac- 429–437. tites. Meteorit. Planet. Sci. 33, 349–359. Floran, R.J., Greive, R.A.F., Phinney, W.C., Warner, J.L., Simonds, C.H., Blanchard, D.P., Melosh, H.J., 1989. Impact cratering — a geological process. Oxf. Monogr. Geol. Geo- Dence, M.R., 1978. Manicouagan Impact Melt, Quebec, 1. Stratigraphy, petrology, phys. 11, 1–245. and chemistry. J. Geophys. Res. 83, 2737–2759. Menneken, M., Nemchin, A.A., Geisler, T., Pidgeon, R.T., Wilde, S.A., 2007. Hadean Froude, D.O., Ireland, T.R., Kinny, P.D., Williams, I.S., Compston, W., 1983. Ion microprobe diamonds in zircon from Jack Hills, Western Australia. Nature 448, 917–920. identification of 4,100–4,200 Myr-old terrestrial zircons. Nature 304, 616–618. Moser, D.E., 1997. Dating the shock wave and thermal imprint of the giant Vredefort Fu, B., Page, Z., Cavosie, A.J., Fournelle, J., Kita, N.T., Lackey, J.S., Wilde, S.A., Valley, J., impact, South Africa. Geology 25, 7–10. 2008. Ti-in-zircon thermometry: applications and limitation. Contrib. Miner. Pet- Neukum, G., Ivanov, B.A., Hartmann, W.K., 2001. Cratering records in the inner solar rol. 156, 197–215. system in relation to the lunar reference system. Space Sci. Rev. 96, 55–86. M.M. Wielicki et al. / Earth and Planetary Science Letters 321-322 (2012) 20–31 31

O'Connel-Cooper, C.D., Spray, J.G., 2011. Geochemistry of the Manicouagan Impact Melt Solomon, S.C., 1980. Differentiation of crusts and cores of the terrestrial planets: les- Sheet: evidence for fractional crystallization. J. Geophys. Res. 116, 1–22. sons for the early Earth? Precambrian Res. 10, 177–194. Paces, J.B., Miller, J.D., 1993. Precise U–Pb ages of Duluth Complex and related maﬁc Taylor, D.J., McKeegan, K.D., Harrison, T.M., 2009. Lu–Hf zircon evidence for rapid lunar intrusions, Northeastern Minnesota: geochronological insights to physical, pet- differentiation. Earth Planet. Sci. Lett. 279, 157–164. rogenetic, paleomagnetic, and tectonomagmatic processes associated with the Tera, F., Papanastassiou, D.A., Wasserburg, G.J., 1974. Isotopic evidence for a terminal 1.1 Ga midcontinent rift system. J. Geophys. Res. 98, 13,997–14,013. lunar cataclysm. Earth Planet. Sci. Lett. 22, 1–21. Pearce, N., Perkins, W., Westgate, J., Gorton, M., Jackson, S., Neal, C., Chenery, S., 1997. A Therriault, A.M., Greive, R.A.F., Reimold, W.U., 1997. Original size of the Vredefort compilation of new and published major and trace element data for NIST SRM 610 Structure: Implications for the geological evolution of the Witwatersrand Basin. and NIST SRM 612 glass reference materials. Geostand. Newslett. 21, 115–144. Meteorit. Planet. Sci. 32, 71–77. Peck, W.H., Valley, J.W., Wilde, S.A., Graham, C.M., 2001. Oxygen isotope ratios and rare Therriault, A.M., Fowler, A.D., Greive, R.A.F., 2002. The Sudbury Igneous Complex: a earth elements in 3.3 to 4.4 Ga zircons: ion microprobe evidence for high δ18O con- differentiated impact melt sheet. Econ. Geol. 97, 1521–1540. tinental crust and oceans in the Early Archean. Geochim. Cosmochim. Acta 65, Watson, E.B., 1996. Surface enrichment and trace-element uptake during crystal 4215–4229. growth. Geochim. Cosmochim. Acta 60, 5013–5020. Poujol,M.,Anhaeusser,C.R.,Armstrong,R.A., 2002. Episodic granitoid emplacement Watson, E.B., Harrison, T.M., 1983. Zircon saturation revisited: temperature and com- in the Archaean Amalia–Kraaipan terrane, South Africa: conﬁrmation from single position effects in a variety of crustal magma types. Earth Planet. Sci. Lett. 64, zircon U–Pb geochronology. J. Afr. Earth Sci. 35, 147–161. 295–304. Reimold, W.U., 1991. Geochemistry of pseudotachylites from the Vredefort Dome, Watson, E.B., Harrison, T.M., 2005. Zircon thermometer reveals minimum melting con- South Africa. Neues Jahrb. Mineral. Abh. 161, 151–184. ditions on earliest Earth. Science 308, 841–844. Reimold, W.U., Gibson, R.L., 1996. Geology and evolution of the Vredefort Impact Struc- Watson, E.B., Harrison, T.M., 2006. Response to Comments on “Zircon thermometer ture, South Africa. J. Afr. Earth Sci. 23, 125–162. reveals minimum melting conditions on earliest Earth”. Science 311, 779. Reimold, W.U., Gibson, R.L., 2006. The melt rocks of the Vredefort impact structure– Wetherill, G.W., 1980. Formation of the terrestrial planets. Annu. Rev. Astron. Astro- Vredefort granophyre and pseudotachylitic breccias: implications for impact phys. 18, 77–113. cratering and the evolution of the Witwatersrand Basin. Chem. Erde 66, 1–35. Wiedenbeck, M., Hanchar, J.M., Peck, W.H., Sylvester, P., Valley, J., Whitehouse, M., Rudnick, R.L., Fountain, D.M., 1995. Nature and composition of the continental crust: a Kronz, A., Morishita, Y., Nasdala, L., 2004. Further characterization of the 91500 lower crustal perspective. Rev. Geophys. 33, 267–309. zircon crystal. Geostand. Geoanal. Res. 28, 9–39. Rudnick, R.L., Gao, S., 2003. Composition of the continental crust. Treatise Geochem. 3, Williams, I.S., 2007. Earth science: old diamonds and the upper crust. Nature 448, 1–64. 880–881. Schmitz, M.D., Bowring, S.A., de Wit, M.J., Gartz, V., 2004. Subduction and terrane colli- Wyllie, P.J., 1977. Crustal anatexis: an experimental review. Tectonophysics 43, 41–71. sion stabilize the western Kaapvaal craton tectosphere 2.9 billion years ago. Earth Zieg, M.J., Marsh, B.D., 2005. The Sudbury Igneous Complex: viscous emulsion differen- Planet. Sci. Lett. 222, 363–376. tiation of a superheated impact melt sheet. Geol. Soc. Am. Bull. 117, 1427–1450. Simonds, C.H., Phinney, W.C., McGee, P.E., Cochran, A., 1978. West Clearwater, Quebec impact structure. I — Field geology, structure and bulk chemistry. II — Petrology. 9th Lunar and Planetary Science Conference, p. 2633.